CN102682779B - Double-channel encoding and decoding method for 3D audio frequency and codec - Google Patents

Abstract

The present invention provides a 3D audio-oriented binaural codec method and a codec. On the basis of the 3D audio-oriented binaural technology, the present invention uses more coding energy for coding according to the auditory characteristics of the human ear. In the main component of the system, different coding methods are used to encode different audio signals, and then a 3D audio-oriented two-channel codec method and corresponding codec are proposed. The method of the invention can reduce codec noise, make the reconstructed signal have a higher signal-to-noise ratio, and can better simulate 3D audio signals at the same time.

Description

Two-channel codec method and codec for 3D audio

technical field

The invention relates to the technical field of audio compression, in particular to a 3D audio-oriented binaural codec method and a codec.

Background technique

With the rapid development of information technology in the new century, audio compression technology has been widely used. Today's 3D audio technologies, such as 5.1 channels, 7.1 channels, and even more channels for sound rendering are becoming more and more popular. Multi-channel audio provides a more immersive and realistic listening experience. However, with the continuous increase of audio channels, the bit rate generated by encoding is also increasing linearly, thus requiring more audio recording space and more real-time transmission bandwidth, so many efficient encoding techniques have emerged as the times require, as follows Mixing parameters Stereo encoding. Meanwhile, many audio codecs for stereo encoding have also been produced for the above-mentioned technologies, such as PS, EAAC+, MPEG-Surround, and stereo audio codecs based on PCA. In the case of multiple sound sources and multiple directions, the codec results of traditional audio codecs cannot show better subjective and objective sound quality.

Contents of the invention

In order to further improve audio codec quality, reduce codec noise, and enhance subjective and objective sound quality, the present invention proposes a 3D audio-oriented two-channel codec method and codec.

In order to solve the problems of the technologies described above, the present invention adopts the following technical solutions:

One, a kind of binaural coding method for 3D audio, comprising steps:

S1.1 performs time-frequency transformation on the input binaural signals respectively, and converts the binaural signals in the time domain into binaural signals in the frequency domain;

S1.2. Dividing the binaural signals in the frequency domain into subbands respectively to obtain binaural subband signals;

S1.3, using the parameter coding method based on the frequency domain principal component and the polar coordinate principal component to encode the two-channel sub-band signals one by one, so as to obtain each two-channel sub-band signal in the above two coding methods The encoding noise energy generated under

ε2.k为第k个双声道子带信号的编码噪音能量，ρ_k(j)为第k个双声道子带信号中第j个频点的信号幅值，

L_k(j)、R_k(j)分别为第k个左声道子带信号和右声道子带信号中第j个频点的信号，n为第k个双声道子带信号中频点的数量；The encoding noise energy obtained by encoding the two-channel subband signal by using the parameter encoding method based on the polar coordinate principal component

ε2.k is the coding noise energy of the kth binaural subband signal, ρ _k (j) is the signal amplitude of the jth frequency point in the kth binaural subband signal,

L _k (j), R _k (j) are the signals of the jth frequency point in the kth left channel subband signal and the right channel subband signal respectively, and n is the intermediate frequency of the kth two-channel subband signal the number of points;

S1.4. For each binaural subband signal, select the parameter encoding method corresponding to the smaller encoding noise energy to further encode the binaural subband signal. If the noise energy is equal, select the parameter encoding method based on the frequency domain principal component The parametric encoding method further encodes the binaural subband signal; if the parameter encoding method based on the frequency domain principal component is used for further encoding, the encoded principal component sequence, direction angle and noise energy ratio of the binaural subband signal are output ; If the parameter encoding method based on the polar coordinate principal component is used for further encoding, the encoded principal component sequence, rotation radius and noise energy ratio of the binaural subband signal are output;

The encoded principal component sequence obtained by adopting the parameter encoding method based on the polar coordinate principal component is:

PC _k ＝{PC _k (j)|j＝1,2,...,n}

表示第k个双声道子带信号中第j个频点的方向角，

L_k(j)、R_k(j)分别为第k个左声道子带信号和右声道子带信号中第j个频点的信号，n为编号为k的子带中频点的数量；Among them, PC _k is the principal component sequence of the kth binaural subband signal, PC _k (j) is the main component of the jth frequency point in the kth binaural subband signal,

Indicates the direction angle of the jth frequency point in the kth binaural subband signal,

L _k (j), R _k (j) are the signals of the jth frequency point in the kth left channel subband signal and the right channel subband signal respectively, and n is the number of mid-frequency points in the subband numbered k ;

The radius of rotation obtained by using the parameter encoding method based on the polar coordinate principal component is:

${\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\sqrt{{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}}$

为第k个双声道子带信号的旋转半径，L_k(j)、R_k(j)分别为第k个左声道子带信号和右声道子带信号中第j个频点的信号，n为第k个双声道子带信号中频点的数量；in,

is the radius of rotation of the kth binaural subband signal, L _k (j) and R _k (j) are the jth frequency points in the kth left channel subband signal and the right channel subband signal respectively signal, n is the number of intermediate frequency points of the kth binaural subband signal;

The noise energy ratio obtained by adopting the parameter coding method based on the polar coordinate principal component is:

$\mathrm{<span\; class="notranslate">\; PAR</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&pi;</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; 48</span>}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; [</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}$

L_k(j)、R_k(j)分别为第k个左声道子带信号和右声道子带信号中第j个频点的信号，n为第k个双声道子带信号中频点的数量；Among them, ρ _k (j) is the signal amplitude of the jth frequency point in the kth binaural subband signal,

L _k (j), R _k (j) are the signals of the jth frequency point in the kth left channel subband signal and the right channel subband signal respectively, and n is the intermediate frequency of the kth two-channel subband signal the number of points;

S1.5. Downmixing the coded principal component sequence to obtain a downmix signal;

S1.6. Encode the downmix signal by using the core encoder to obtain an encoded code stream, and write the direction angle or rotation radius and noise energy ratio into the encoded code stream.

Two, a kind of binaural encoder for 3D audio, comprising:

A time-frequency conversion module is used to perform time-frequency conversion on the input binaural signals respectively, and convert the binaural signals in the time domain into binaural signals in the frequency domain;

The sub-band division module is used to divide the binaural signals in the frequency domain into sub-bands respectively to obtain binaural sub-band signals;

ε_2.k为第k个双声道子带信号的编码噪音能量，ρ_k(j)为第k个双声道子带信号中第j个频点的信号幅值，

L_k(j)、R_k(j)分别为第k个左声道子带信号和右声道子带信号中第j个频点的信号，n为第k个双声道子带信号中频点的数量；The coding noise energy calculation module is used to encode the two-channel sub-band signals one by one by adopting the parameter coding method based on the frequency-domain principal component and the polar-coordinate principal component, so as to obtain each two-channel sub-band signal in the above-mentioned Encoding noise energy produced under two encoding methods; the encoding noise energy obtained by encoding the two-channel subband signal by using the parameter encoding method based on polar coordinate principal components

ε _2.k is the coding noise energy of the kth binaural subband signal, ρ _k (j) is the signal amplitude of the jth frequency point in the kth binaural subband signal,

L _k (j), R _k (j) are the signals of the jth frequency point in the kth left channel subband signal and the right channel subband signal respectively, and n is the intermediate frequency of the kth two-channel subband signal the number of points;

The parameter coding module is used for each binaural sub-band signal, and selects the parameter coding method corresponding to the smaller coding noise energy to further encode the binaural sub-band signal. The parametric coding method of the component is used to further encode the binaural subband signal; if the parameter coding method based on the frequency domain principal component is used for further coding, the encoded principal component sequence, direction angle and noise of the binaural subband signal are output Energy ratio; if the parameter encoding method based on the polar coordinate principal component is used for further encoding, the encoded principal component sequence, rotation radius and noise energy ratio of the binaural subband signal are output;

The encoded principal component sequence obtained by adopting the parameter encoding method based on the polar coordinate principal component is:

PC _k ＝{PC _k (j)|j＝1,2,...,n}

表示第k个双声道子带信号中第j个频点的方向角，

L_k(j)、R_k(j)分别为第k个左声道子带信号和右声道子带信号中第j个频点的信号，n为编号为k的子带中频点的数量；Among them, PC _k is the principal component sequence of the kth binaural subband signal, PC _k (j) is the main component of the jth frequency point in the kth binaural subband signal,

Indicates the direction angle of the jth frequency point in the kth binaural subband signal,

L _k (j), R _k (j) are the signals of the jth frequency point in the kth left channel subband signal and the right channel subband signal respectively, and n is the number of mid-frequency points in the subband numbered k ;

The radius of rotation obtained by using the parameter encoding method based on the polar coordinate principal component is:

${\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\sqrt{{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}}$

为第k个双声道子带信号的旋转半径，L_k(j)、R_k(j)分别为第k个左声道子带信号和右声道子带信号中第j个频点的信号，n为第k个双声道子带信号中频点的数量；in,

is the radius of rotation of the kth binaural subband signal, L _k (j) and R _k (j) are the jth frequency points in the kth left channel subband signal and the right channel subband signal respectively signal, n is the number of intermediate frequency points of the kth binaural subband signal;

The noise energy ratio obtained by adopting the parameter coding method based on the polar coordinate principal component is:

$\mathrm{<span\; class="notranslate">\; PAR</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&pi;</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; 48</span>}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; [</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}$

L_k(j)、R_k(j)分别为第k个左声道子带信号和右声道子带信号中第j个频点的信号，n为第k个双声道子带信号中频点的数量；Among them, ρ _k (j) is the signal amplitude of the jth frequency point in the kth binaural subband signal,

L _k (j), R _k (j) are the signals of the jth frequency point in the kth left channel subband signal and the right channel subband signal respectively, and n is the intermediate frequency of the kth two-channel subband signal the number of points;

A downmix module, used to downmix the coded principal component sequence to obtain a downmix signal;

The core encoder is used to encode the downmix signal to obtain a coded code stream, and write the direction angle or rotation radius and noise energy ratio into the coded code stream.

Three, a kind of binaural decoding method for 3D audio, comprising steps:

S2.1 Use the core decoder to decode the encoded code stream to obtain the decoded signal;

S2.2 Divide the decoded signal into subbands to obtain decoded subband signals;

S2.3 Use the parameter decoding method corresponding to the parameter encoding method used for encoding, and combine the direction angle or rotation radius in the encoded code stream, and the noise capability ratio to decode the decoded subband signal to obtain the reconstructed frequency domain subband Signal;

S2.4 Merging the reconstructed frequency-domain sub-band signals to obtain a reconstructed frequency-domain signal;

S2.5 Perform time-frequency inverse transformation on the frequency domain signal, convert the frequency domain signal into a time domain signal, and restore the reconstructed audio signal.

The above parameter decoding method is a parameter decoding method based on frequency domain principal components or a parameter decoding method based on polar coordinate principal components.

The described decoded subband signal is decoded by using the parameter decoding method based on the frequency domain principal component to obtain the reconstructed frequency domain subband signal, specifically: according to the noise energy ratio in the coded code stream, a The signal has white noise with the same energy, combined with the main component sequence and direction angle in the encoded code stream, the decoded sub-band signal is restored to obtain the reconstructed frequency-domain sub-band signal.

The described decoded subband signal is decoded by using the parametric decoding method based on the polar coordinate principal component to obtain a reconstructed frequency domain subband signal, specifically: according to the noise energy ratio in the coded code stream, a The signal has white noise with the same energy, combined with the main component sequence and rotation radius in the encoded code stream, the decoded sub-band signal is restored to obtain the reconstructed frequency-domain sub-band signal.

Four, a kind of two-channel decoder facing 3D audio, comprising:

The core decoder is used to decode the encoded code stream to obtain the decoded signal;

A subband division module, used to divide the decoded signal into subbands to obtain decoded subband signals;

The parameter decoding module is used to decode the decoded sub-band signal by using a parameter decoding method corresponding to the parameter encoding method used for encoding, and combining the direction angle or rotation radius and noise capability ratio in the encoded code stream to obtain the reconstructed frequency domain subband signal;

A subband combining module, used to combine the reconstructed frequency domain subband signals to obtain a reconstructed frequency domain signal;

The time-frequency inverse transform module is used to perform time-frequency inverse transform on the frequency domain signal, convert the frequency domain signal into a time domain signal, and restore the reconstructed audio signal.

The above parameter decoding module further includes a parameter decoding module based on frequency domain principal components and a parameter decoding module based on polar coordinate principal components.

The parameter decoding module based on frequency-domain principal components is used to generate a white noise with the same energy as the original signal according to the noise energy ratio in the coded code stream, combined with the main component sequence and direction angle in the coded code stream, The decoded sub-band signal is restored to obtain a reconstructed frequency-domain sub-band signal.

The parameter decoding module based on the polar coordinate principal component is used to generate a white noise having the same energy as the original signal according to the noise energy ratio in the encoded code stream, combined with the principal component sequence and the radius of rotation in the encoded code stream, The decoded sub-band signal is restored to obtain a reconstructed frequency-domain sub-band signal.

Based on the 3D audio-oriented binaural technology, the present invention uses more coding energy in the main component of coding according to the auditory characteristics of the human ear, and uses different coding methods for different audio signals to code, and then proposes A binaural codec method for 3D audio and a corresponding codec. The method of the invention can reduce codec noise, make the reconstructed signal have a higher signal-to-noise ratio, and can better simulate 3D audio signals at the same time.

Description of drawings

Fig. 1 is the flow chart of coding method of the present invention;

Fig. 2 is the flowchart of decoding method of the present invention;

Fig. 3 is the flowchart of subband division in the encoding method of the present invention;

Fig. 4 is the flowchart of coding method selection in the coding method of the present invention;

Fig. 5 is a schematic diagram of the parameter encoding method based on the polar coordinate principal component of the present invention;

Fig. 6 is a flowchart of decoding method selection in the decoding method of the present invention;

Fig. 7 is a flowchart of parameter decoding in the decoding method of the present invention.

Detailed ways

The present invention proposes a binaural encoding method for 3D audio and a corresponding binaural decoding method. During specific implementation, those skilled in the art can use computer software means to realize automatic audio encoding and decoding according to the provided technical solution. Since in the codec application, the codec software method can often be solidified to form a codec device, so the present invention also provides a corresponding 3D audio-oriented two-channel encoder and decoder.

The specific implementation manners of the present invention will be described in detail below in conjunction with the accompanying drawings, so as to make the technical solutions and beneficial effects of the present invention clearer.

In the present invention, in order to use the parameter coding method based on the frequency domain principal component to analyze the spatial audio signal, the minimum mean square error (MMSE) is used in the coding scheme to combine the two channels into one channel, and then only this one channel is used by the core encoded by the encoder. During decoding, the signal is reconstructed using the orientation angle, the ambient noise power ratio (PAR) of the principal component and the secondary component, where the ambient noise produces a white noise similar to the energy of the original signal to simulate the original signal. However, for a 3D multi-channel signal, when sub-bands are divided, some sub-bands are merged from small uniform sub-bands, which include many sub-bands with different energy ratios between left and right channels. Since these subbands can better simulate multiple sound sources in different directions, it is unreasonable to use only one direction angle and PAR to transmit the downmixed channel in the parametric coding method based on the frequency domain principal component. In view of the above problems, the present invention proposes a parameter encoding method based on polar coordinates, in which the parameter encoding of the main component and the secondary component is performed in polar coordinates, and the signal is reconstructed with the radius of rotation and PAR to better simulate 3D audio signal to have a higher signal-to-noise ratio.

The two-channel encoding method for 3D audio of the present invention, the specific flow chart is referring to Fig. 1, comprises the following steps:

Step 1.1, perform time-frequency transformation on the input binaural signal respectively, and convert the binaural signal in the time domain into a binaural signal in the frequency domain

The two-channel signal is composed of the left channel signal l and the right channel signal r. The specific implementation of this step is: using the fast Fourier transform (FFT) to transform the left channel signal l and the right channel signal r in the time domain They are respectively converted into the left channel signal L and the right channel signal R in the frequency domain.

In step 1.2, divide the left channel signal L and the right channel signal R into subbands in the frequency domain to obtain left and right channel subband signals. FIG. 3 is a flow chart of a specific implementation of this step.

The specific implementation of this step is:

The left channel signal L and the right channel signal R in the frequency domain are divided into 64 subbands by using the division method based on the equivalent rectangular bandwidth (ERB). The subbands of the channel signal L and the right channel signal R are combined or subdivided, or both are combined and subdivided to obtain the final subband signal of the left channel and the right channel signal.

Since the human ear is more sensitive to low-frequency sounds, but less sensitive to high-frequency sounds, the 64 sub-bands of the left channel signal L and the right channel signal R can be further processed: the low-frequency sub-bands can be processed Subdividing the subbands, or merging the high frequency subbands, or both subdividing the low frequency subbands and merging the high frequency subbands. In this specific implementation, the 3 low-frequency sub-bands of the 64 sub-band signals are subdivided into 16 sub-bands, and the 61 high-frequency sub-bands are merged into 4 sub-bands, and finally 20 sub-band signals are obtained. The following operations are aimed at The resulting 20 subband signals are performed. The ranges of the above-mentioned low frequency and high frequency are artificially defined according to needs during specific implementation.

In step 1.3, the left channel sub-band signal and the right channel sub-band signal obtained in step 1.2 are respectively processed by the parametric coding method (PCA) based on the frequency domain principal component and the parametric coding method (PC-PCA) based on the polar coordinate principal component Encoding is carried out, and the encoding noise energies of the above two parameter encoding methods are obtained respectively.

The specific implementation of this step is:

1) The left channel subband signal and the right channel subband signal are encoded by the parametric coding method based on the frequency domain principal component to obtain the coding noise energy generated by the parametric coding method based on the frequency domain principal component.

Assuming that the number of left channel subband signal L _k and right channel subband signal R _k obtained in step 1.2 is N, the kth left channel subband signal and right channel subband signal are expressed as L _k , R _k , k=1, 2,..., N, and assuming that the left channel sub-band signal L _k and the right channel sub-band signal R _k each contain n frequency points, then the sub-band signals L _k and R _k can be regarded as a sequence composed of signals of n frequency points, L _k ={L _k (j)|j=1,2,...,n} and R _k ={R _k (j)| j=1,2,...,n}, L _k (j) and R _k (j) are the signals of the jth frequency point in the subband signals L _k and R _k respectively. This step is to obtain the coding noise energy generated by the parameter coding method based on the frequency domain principal component for each sub-band signal L _k and R _k one by one, k=1, 2, . . . , N.

The following will take sub-band signals L _k and R _k as examples to further illustrate the acquisition of coding noise energy generated by the parameter coding method based on frequency domain principal components:

a) Calculate the covariance matrix R _k composed of L _k and R _k sequences:

${\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\left[\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; ll</span>}}& {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; lr</span>}}\\ {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; rl</span>}}& {\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; rr</span>}}\end{array}\right]<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

in,

r _ll =cov[L _k ,L _k ], r _lr =r _rl =cov[L _k ,R _k ], r _rr =cov[R _k ,R _k ];

b) Find the eigenvalues λ ₁ and λ ₂ of the covariance matrix R _k :

${\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; [</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; ll</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; rr</span>}}<span\; class="notranslate">\; +</span>\sqrt{{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; ll</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; rr</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; lr</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

${\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; [</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; ll</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; rr</span>}}<span\; class="notranslate">\; +</span>\sqrt{{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; ll</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; rr</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; lr</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

c) According to the eigenvalues λ ₁ and λ ₂ , the main component energy E _p and the secondary component energy E _s of the parameter coding method (PCA) based on the frequency domain principal components are obtained respectively:

E _p ＝max(λ ₁ ,λ ₂ ) (4)

E _s =min(λ ₁ ,λ ₂ ) (5)

Then, the encoding noise energy ε ₁ =E _s =min(λ ₁ ,λ ₂ ) generated by the parameter encoding method based on the frequency domain principal components.

2) The left channel sub-band signal and the right channel sub-band signal are encoded by the parametric coding method based on the polar coordinate principal component, so as to obtain the coding noise energy generated by the parametric coding method based on the polar coordinate principal component.

The parameter coding method based on the polar coordinate principal component is self-created on the basis of the frequency domain principal component parameter coding method. The coding principles of the two are the same, but the coordinates used are different. The parameter coding method based on the frequency domain principal component is a rectangular coordinate system, while the parameter encoding method based on the polar coordinate principal component uses a polar coordinate system.

Assuming that the number of left channel subband signal L _k and right channel subband signal R _k obtained in step 1.2 is N, the kth left channel subband signal and right channel subband signal are expressed as L _k , R _k , k=1,2,...,N, and assuming that the left channel sub-band signal L _k and the right channel sub-band signal R _k contain n frequency points, then the sub-band signals L _k and R _k can be seen as a sequence composed of signals at n frequency points, L _k ={L _k (j)|j=1,2,...,n} and R _k ={R _k (j)|j =1,2,...,n}, L _k (j) and R _k (j) are the signals of the jth frequency point in the subband signals Lk and Rk respectively. This step is to obtain the encoding noise energy generated by the parameter encoding method based on the polar coordinate principal component for each sub-band signal L _k and R _k one by one, k=1, 2, . . . , N.

The following will take the sub-band signals L _k and R _k as examples to further illustrate this step:

如图5所示，其中，j＝1,2,...,n，L_k(j)、Rk(j)表示子带信号L_k和R_k中第j个频点的信号，ρ_k(j)表示子带信号Lk和Rk中第j个频点的信号的幅值，

表示子带信号L_k和R_k中第j个频点的方向角：a) In order to encode the principal component parameters in the polar coordinate system, the sub-band signals L _k (j) and Rk (j) of each frequency point in the sub-band signals L _k and R _k are introduced into the polar coordinate system one by one to form 2 a new random variable ρ _k (j) and

As shown in Figure 5, where j=1, 2,..., n, L _k (j), Rk (j) represent the signal of the jth frequency point in the sub-band signals L _k and R _k , ρ _k (j) represents the amplitude of the signal of the jth frequency point in the sub-band signals Lk and Rk,

Indicates the direction angle of the jth frequency point in the subband signals L _k and R _k :

序列：The signal amplitudes of each frequency point in the sub-band signals L _k and R _k form a ρ _k sequence, and the direction angles corresponding to each frequency point in the sub-band signals L _k and R _k form

sequence:

ρ _k ＝{ρ _k (j)|j＝1,2,...,n} (7)

序列构成的协方差矩阵R_k，b) Calculate ρ _k and

The covariance matrix R _k composed of sequences,

in,

和次要成分能量E_ρ：c) Calculate the eigenvalues λ ₁ and λ ₂ of the covariance matrix R _k (9), and obtain the principal component energy of the parameter encoding method (PC-PCA) based on the polar coordinate principal component based on λ ₁ and λ ₂

and the minor component energy E _ρ :

${\mathrm{<span\; class="notranslate">\; E.</span>}}_{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; [</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; ]</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 11</span>}<span\; class="notranslate">\; )</span>$

Then, the encoding noise energy ε ₂ =E _ρ of the parameter encoding method based on the polar coordinate principal component.

The above parametric coding methods based on frequency-domain principal components and polar coordinate principal components are used to solve the coding noise energy for the N subband signals Lk and Rk one by one, and finally N groups of coding noise energies are obtained.

Step 1.4: Select the optimal parameter coding method according to the coding noise energy generated by the above two parameter coding methods, and use the selected parameter coding method to process the left and right channel subband signals (L _k and R _k ) for further encoding

The specific implementation of selecting the optimal parameter encoding method in this step is as follows:

Select a parameter encoding method with less coding noise energy, and output the mode mode corresponding to the parameter encoding method, and then use the selected parameter encoding method to further encode the left and right channel signals obtained in step 1.2.

Assuming that the coding noise energy generated by encoding the subband signals Lk and Rk by using the parameter coding method based on the frequency domain principal component and the polar coordinate principal component is ε ₁ , ε ₂ , the subband signals L _k and R _k are still taken as An example to illustrate the specific implementation of this step:

1) If ε ₁ ≤ ε ₂ , then output mode=0. At this time, the sub-band signals L _k and R _k are further encoded by the parameter encoding method based on the frequency domain principal component:

According to the covariance matrix R _k of formula (1), the direction angles of the subband signals L _k and R _k are obtained

The sub-band signals L _k and R _k are further coded by the parameter coding method based on frequency domain principal components, and the coded principal component sequence PC _k and secondary component sequence A _k are obtained, PC _k = {PC _k (j)| j=1,2,...,n}, A _k ={A _k (j)|j=1,2,...,n}, PC _k (j) is the sub-band signals L _k and R _k The main component of the jth frequency point in A _k (j) is the secondary component of the jth frequency point in the subband signals L _k and R _k , where:

$\left(\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}& \mathrm{<span\; class="notranslate">\; sin</span>}{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}\\ <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; sin</span>}{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}& \mathrm{<span\; class="notranslate">\; cos</span>}{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}\end{array}\right)\left(\begin{array}{c}{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>\\ {\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>\end{array}\right)<span\; class="notranslate">\; =</span>\left(\begin{array}{c}{\mathrm{<span\; class="notranslate">\; PC</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>\\ {\mathrm{<span\; class="notranslate">\; A</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>\end{array}\right)<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 12</span>}<span\; class="notranslate">\; )</span>$

L _k (j) and R _k (j) are the signals of the jth frequency point in the sub-band signals L _k and R _k respectively, θ _k represents the direction angle of the sub-band signals L _k and R _k , k=1,2 ,...,N, j=1,2,...,n.

All subbands are decoded one by one by the above method, and the main component sequence PC _k , direction angle θ _k , and noise energy ratio PAR (that is, the ratio of E _p to E _s ) of each subband are output.

2) If ε ₁ >ε ₂ , then output mode=1. At this time, the sub-band signals L _k and R _k are further encoded using the parameter encoding method based on the polar coordinate principal component:

The sub-band signals L _k and R _k are further encoded by the parameter encoding method based on the polar coordinate principal component, and the encoded principal component sequence PC _k and secondary component sequence A _k are obtained, PC _k = {PC _k (j)| j=1,2,...,n}, A _k ={A _k (j)|j=1,2,...,n}, PC _k (j) is the sub-band signals L _k and R _k The main component of the jth frequency point in A _k (j) is the secondary component of the jth frequency point in the subband signals L _k and R _k :

表示子带信号中L_k和R_k第j个频点的方向角，

的值如式（6）所示，k＝1,2,...,N，j＝1,2,...,n。Among them, L _k (j) and R _k (j) are the signals of the jth frequency point in the sub-band signals L _k and R _k respectively,

Indicates the direction angle of the jth frequency point of L _k and R _k in the subband signal,

The value of is shown in formula (6), k=1,2,...,N, j=1,2,...,n.

旋转半径

为子带信号L_k和R_k各频点的信号幅值的平均值，即：Find the radius of rotation for the subband signals L _k and R _k

radius of rotation

is the average value of the signal amplitudes of each frequency point of the sub-band signals L _k and R _k , that is:

${\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\sqrt{{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 14</span>}<span\; class="notranslate">\; )</span>$

之比）。Use the above method to decode all subbands one by one, and output the main component sequence PC _k , rotation radius ρ _k , and PAR (ie E _ρ and

Ratio).

Steps 1.3 and 1.4 are encoded based on each sub-band signal. For each sub-band signal, a coding noise energy ε ₁ based on the frequency-domain principal component-based parameter coding method and a polar-coordinate principal component-based For the coding noise energy ε ₂ of the parametric coding method, each sub-band signal is compared with ε ₁ and ε ₂ once, and the parameter coding method corresponding to the smaller coding noise energy is selected to further code the sub-band. The process of steps 1.3 and 1.4 is shown in Figure 3.

Step 1.5, performing signal down-mixing on all principal component sequences PC _k generated in step 1.4, to obtain down-mixed signal m, k=1,2,...,N;

In step 1.6, pass the downmix signal m obtained in step 1.5 into the core encoder for encoding, and obtain the encoded code stream. If the parameter encoding method based on the polar coordinate principal component is used for encoding, the rotation radius ρ _k , PAR and mode values are written into the coded code stream; if the parameter coding method based on the principal component in the extreme frequency domain is used for coding, the direction angle θ _k , PAR and mode values are written into the coded code stream.

The present invention also provides a two-channel encoding method for 3D audio, the specific flow chart is shown in Figure 2, including the following steps:

Step 2.1, decode the encoded code stream obtained by the encoding end to obtain the decoded signal m

During specific implementation, the coded code stream is input into the core decoder, and the core decoder is used to decode to obtain the decoded signal m.

Step 2.2, divide the decoded signal m obtained in step 2.1 into subbands to obtain decoded subband signals

During specific implementation, the decoded signal m output by the core decoder is divided into subband sequences P(N), where N is the number of subbands, which is equivalent to the N value in the encoding method.

Step 2.3, select the corresponding decoding mode according to the mode mode value in the encoded code stream, and perform decoding work in combination with the direction angle or rotation radius and noise energy ratio in the encoded code stream to obtain the reconstructed frequency domain subband signal, as shown in Figure 6 and Figure 7 shows.

The specific implementation of this step is:

1) If mode=0, select the parameter decoding method based on the frequency domain principal component:

和According to the noise energy ratio PAR in the encoded code stream, a white noise with the same energy as the original signal is generated, combined with the main component sequence and direction angle in the encoded code stream, the parameter decoding method based on the frequency domain principal component is used to convert the The obtained subband sequence P(N) is restored to obtain the decoded subband signal, that is, the reconstructed frequency domain subband signal

and

${\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; L</span>}}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; L</span>}}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>{\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; L</span>}}}_{\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; .</span>$

2) If mode=1, select the parameter decoding method based on the polar coordinate principal component:

According to the noise energy ratio PAR in the encoded code stream, a white noise with the same energy as the original signal is generated, combined with the main component sequence and rotation radius in the encoded code stream, the parameter decoding method based on the polar coordinate principal component is used to convert the The obtained subband sequence P(N) is restored to obtain the decoded subband signal, that is, the reconstructed frequency domain subband signal ${\hat{R}}_1,{\hat{R}}_2,...,{\hat{R}}_N.$ and ${\hat{L}}_1,{\hat{L}}_2,...,{\hat{L}}_N.$

和

Step 2.4, combining the reconstructed frequency-domain sub-band signals obtained in step 2.3 to obtain a reconstructed frequency-domain signal

and

和具体实施时，可采用现有技术，如FFT（快速傅里叶变换）变换来实现，本发明不予赘述。Step 2.5, performing time-frequency inverse transform on the reconstructed frequency-domain channel signal obtained in step 2.4, recovering the reconstructed time-domain signal

and During specific implementation, existing technologies such as FFT (Fast Fourier Transform) transformation can be used for implementation, which will not be described in detail in the present invention.

The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which the present invention belongs can make various modifications or supplements to the described specific embodiments or adopt similar methods to replace them, but they will not deviate from the spirit of the present invention or go beyond the definition of the appended claims range.

Claims (10)

1. A binaural encoding method for 3D audio, characterized in that, comprising steps: S1.1. Perform time-frequency transformation on the input binaural signals respectively, and convert the binaural signals in the time domain into binaural signals in the frequency domain; S1.2. Dividing the binaural signals in the frequency domain into subbands respectively to obtain binaural subband signals; S1.3, using the parameter coding method based on the frequency domain principal component and the polar coordinate principal component to encode the two-channel sub-band signals one by one, so as to obtain each two-channel sub-band signal in the above two coding methods The encoding noise energy generated under The encoding noise energy obtained by encoding the two-channel subband signal by using the parameter encoding method based on the polar coordinate principal component ε _2.k is the coding noise energy of the kth binaural subband signal, ρ _k (j) is the signal amplitude of the jth frequency point in the kth binaural subband signal,

R _k (j) is the signal of the jth frequency point in the kth left channel subband signal and the right channel subband signal respectively, and n is the quantity of the frequency point in the kth two-channel subband signal; S1.4. For each binaural subband signal, select the parameter encoding method corresponding to the smaller encoding noise energy to further encode the binaural subband signal. If the noise energy is equal, select the parameter encoding method based on the frequency domain principal component The parametric encoding method further encodes the binaural subband signal; if the parameter encoding method based on the frequency domain principal component is used for further encoding, the encoded principal component sequence, direction angle and noise energy ratio of the binaural subband signal are output ; If the parameter encoding method based on the polar coordinate principal component is used for further encoding, the encoded principal component sequence, rotation radius and noise energy ratio of the binaural subband signal are output; The encoded principal component sequence obtained by adopting the parameter encoding method based on the polar coordinate principal component is: PC _k ={PC _k (j)|j=1,2,...,n} Among them, PC _k is the principal component sequence of the kth binaural subband signal, PC _k (j) is the main component of the jth frequency point in the kth binaural subband signal,

Indicates the direction angle of the jth frequency point in the kth binaural subband signal,

L _k (j), R _k (j) are the signals of the jth frequency point in the kth left channel subband signal and the right channel subband signal respectively, and n is the number of mid-frequency points in the subband numbered k ; The radius of rotation obtained by using the parameter encoding method based on the polar coordinate principal component is: $\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}<span\; class="notranslate">\; =</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\sqrt{{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}}$ in,

is the radius of rotation of the kth binaural subband signal, L _k (j) and R _k (j) are the jth frequency points in the kth left channel subband signal and the right channel subband signal respectively signal, n is the number of intermediate frequency points of the kth binaural subband signal; The noise energy ratio obtained by adopting the parameter coding method based on the polar coordinate principal component is: $\mathrm{<span\; class="notranslate">\; PAR</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&pi;</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; 48</span>}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; [</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}$ Among them, ρ _k (j) is the signal amplitude of the jth frequency point in the kth binaural subband signal,

L _k (j), R _k (j) are the signals of the jth frequency point in the kth left channel subband signal and the right channel subband signal respectively, and n is the intermediate frequency of the kth two-channel subband signal the number of points; S1.5. Downmixing the coded principal component sequence to obtain a downmix signal; S1.6. Encode the downmix signal by using the core encoder to obtain an encoded code stream, and write the direction angle or rotation radius and noise energy ratio into the encoded code stream. 2. A dual-channel encoder for 3D audio, characterized in that it comprises: A time-frequency conversion module is used to perform time-frequency conversion on the input binaural signals respectively, and convert the binaural signals in the time domain into binaural signals in the frequency domain; The sub-band division module is used to divide the binaural signals in the frequency domain into sub-bands respectively to obtain binaural sub-band signals; The coding noise energy calculation module is used to encode the two-channel sub-band signals one by one by adopting the parameter coding method based on the frequency-domain principal component and the polar-coordinate principal component, so as to obtain each two-channel sub-band signal in the above-mentioned Encoding noise energy produced under two encoding methods; the encoding noise energy obtained by encoding the two-channel subband signal by using the parameter encoding method based on polar coordinate principal components

ε _2.k is the coding noise energy of the kth binaural subband signal, ρ _k (j) is the signal amplitude of the jth frequency point in the kth binaural subband signal,

L _k (j), R _k (j) are the signals of the jth frequency point in the kth left channel subband signal and the right channel subband signal respectively, and n is the intermediate frequency of the kth two-channel subband signal the number of points; The parameter coding module is used for each binaural sub-band signal, and selects the parameter coding method corresponding to the smaller coding noise energy to further encode the binaural sub-band signal. The parametric coding method of the component is used to further encode the binaural subband signal; if the parameter coding method based on the frequency domain principal component is used for further coding, the encoded principal component sequence, direction angle and noise of the binaural subband signal are output Energy ratio; if the parameter encoding method based on the polar coordinate principal component is used for further encoding, the encoded principal component sequence, rotation radius and noise energy ratio of the binaural subband signal are output; The encoded principal component sequence obtained by adopting the parameter encoding method based on the polar coordinate principal component is: PC _k ={PC _k (j)|j=1,2,...,n} Among them, PC _k is the principal component sequence of the kth binaural subband signal, PC _k (j) is the main component of the jth frequency point in the kth binaural subband signal,

Indicates the direction angle of the jth frequency point in the kth binaural subband signal,

L _k (j), R _k (j) are the signals of the jth frequency point in the kth left channel subband signal and the right channel subband signal respectively, and n is the number of mid-frequency points in the subband numbered k ; The radius of rotation obtained by using the parameter encoding method based on the polar coordinate principal component is: $\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}<span\; class="notranslate">\; =</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\sqrt{{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}}$ in,

is the radius of rotation of the kth binaural subband signal, L _k (j) and R _k (j) are the jth frequency points in the kth left channel subband signal and the right channel subband signal respectively signal, n is the number of intermediate frequency points of the kth binaural subband signal; The noise energy ratio obtained by adopting the parameter coding method based on the polar coordinate principal component is: $\mathrm{<span\; class="notranslate">\; PAR</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; \&pi;</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; 48</span>}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; [</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; no</span>}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}$ Among them, ρ _k (j) is the signal amplitude of the jth frequency point in the kth binaural subband signal,

L _k (j), R _k (j) are the signals of the jth frequency point in the kth left channel subband signal and the right channel subband signal respectively, and n is the intermediate frequency of the kth two-channel subband signal the number of points; A downmix module, used to downmix the coded principal component sequence to obtain a downmix signal; The core encoder is used to encode the downmix signal to obtain a coded code stream, and write the direction angle or rotation radius and noise energy ratio into the coded code stream. 3. A binaural decoding method for 3D audio, characterized in that it comprises steps: S2.1. Using a core decoder to decode the coded code stream obtained by using the coding method according to claim 1 to obtain a decoded signal; S2.2. Dividing the decoded signal into subbands to obtain decoded subband signals; S2.3. Using the parameter decoding method corresponding to the parameter encoding method used for encoding, and combining the direction angle or rotation radius in the encoded code stream, and the noise capability ratio to decode the decoded sub-band signal, obtain the reconstructed frequency domain sub-band signal with signal; S2.4. Merging the reconstructed frequency-domain sub-band signals to obtain a reconstructed frequency-domain signal; S2.5. Perform time-frequency inverse transformation on the frequency domain signal, convert the frequency domain signal into a time domain signal, and recover the reconstructed audio signal. 4. the binaural decoding method facing 3D audio according to claim 3, characterized in that: The parameter decoding method described in step S2.3 is a parameter decoding method based on frequency domain principal components or a parameter decoding method based on polar coordinate principal components. 5. the two-channel decoding method facing 3D audio according to claim 4, is characterized in that: The described decoded subband signal is decoded by using the parameter decoding method based on the frequency domain principal component to obtain a reconstructed frequency domain subband signal, specifically: According to the noise energy ratio in the encoded code stream, a white noise with the same energy as the original signal is generated, combined with the main component sequence and direction angle in the encoded code stream, the decoded sub-band signal is restored to obtain the reconstructed frequency domain subband signal. 6. the two-channel decoding method facing 3D audio according to claim 4, is characterized in that: The described decoded subband signal is decoded by using the parameter decoding method based on the polar coordinate principal component to obtain a reconstructed frequency domain subband signal, specifically: According to the noise energy ratio in the encoded code stream, a white noise with the same energy as the original signal is generated, combined with the main component sequence and rotation radius in the encoded code stream, the decoded sub-band signal is restored to obtain the reconstructed frequency domain subband signal. 7. A two-channel decoder for 3D audio, characterized in that it comprises: A core decoder, used to decode the encoded code stream obtained by using the encoding method as claimed in claim 1, to obtain a decoded signal; A sub-band division module is used to divide the decoded signal into sub-bands to obtain decoded sub-band signals; The parameter decoding module is used to decode the decoded sub-band signal by using a parameter decoding method corresponding to the parameter encoding method used for encoding, and combining the direction angle or rotation radius and noise capability ratio in the encoded code stream to obtain the reconstructed frequency domain subband signal; A subband combining module, used to combine the reconstructed frequency domain subband signals to obtain a reconstructed frequency domain signal; The time-frequency inverse transform module is used to perform time-frequency inverse transform on the frequency domain signal, convert the frequency domain signal into a time domain signal, and restore the reconstructed audio signal. 8. the two-channel decoder facing 3D audio according to claim 7, is characterized in that: The parameter decoding module further includes a parameter decoding module based on frequency domain principal components and a parameter decoding module based on polar coordinate principal components. 9. the two-channel decoder facing 3D audio according to claim 8, is characterized in that: The parameter decoding module based on frequency-domain principal components is used to generate a white noise with the same energy as the original signal according to the noise energy ratio in the coded code stream, combined with the main component sequence and direction angle in the coded code stream, The decoded sub-band signal is restored to obtain a reconstructed frequency-domain sub-band signal. 10. the two-channel decoder facing 3D audio according to claim 8, is characterized in that: The parameter decoding module based on the polar coordinate principal component is used to generate a white noise having the same energy as the original signal according to the noise energy ratio in the encoded code stream, combined with the principal component sequence and the radius of rotation in the encoded code stream, The decoded sub-band signal is restored to obtain a reconstructed frequency-domain sub-band signal.

Images